While checking for spelling errors major factors are:

• Language Model ie unigram , bigram
• Edit distance and confusion matrix
• Other errors like phonetics and space
• Proper smoothing for words not present in corpus

• Preparation
• First I pruned the given corpus to get the unigram frequency of each words and then also removed the words with wrong spellings.
• I created a script for constructing the confusion matrix from the wrong words.
• Unigram
• Without phonetics error(only confusion matrix)

This gave an less accurate results.
• With phonetics error taken into consideration
The accuracy improved many folds.
• Three Top Suggestions for each word are also given
• Bigram
• With phonetics error taken into consideration
• Three Top Suggestions for each word are also given

The python files used are code unigram
The result files can be seen at: data unigram files

The python files used are code bigram
The result files can be seen at: data bigram files(can't be inserted as disk quota exceeded)

The python file to run aspell file